Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: The superior Rybka chess knowledge

Author: Ross Boyd

Date: 00:26:09 01/19/06

Go up one level in this thread


On January 18, 2006 at 15:51:42, Uri Blass wrote:

>On January 18, 2006 at 15:30:03, Ross Boyd wrote:
>
>>On January 18, 2006 at 15:05:36, Dann Corbit wrote:
>>
>>>On January 18, 2006 at 14:39:40, Fernando Villegas wrote:
>>>
>>>>Some days ago I talked of knowledege to know which knowledge is irrelevant as a
>>>>probable feature of Rybka. And almost nobody answered or commented my post.
>>>>Surely it is an advantage to be a chess programmer in a site like this.
>>>
>>>Probably the only one who can answer correctly is the author.
>>>
>>>But you can look at some interesting generalizations.
>>>
>>>I have found that some evaluation terms add nothing either to the EPD solution
>>>count or the contest point gathering effectiveness of an engine.
>>>
>>>An example is queen mobility.  I don't know of any logical reason why its
>>>computation should not help, but it doesn't.
>>>
>>>I have seen other chess engine authors come to the same conclusion.
>>
>>Agreed. I do pseudo mobility only for bishops and knights. Queen mobility
>>improves nothing. However, I do piece-square so the queen knows about
>>centralisation.
>
>For me it is the opposite about queen.
>
>I do not have piece square table for centralization of the queen but I have
>mobility for all pieces.
>
>I have piece square table for the queen but it is not about centralization and I
>do not know if it is productive(probably I will need a lot of games to test it
>because my piece square table only change the value of queen by at most 0.1 pawn
>and it does not change the value of queen in most squares).
>
>Here is my piece square table for the queen(a8 a7 a2 a1 h8 h7 h2 h1 considered
>to be relatively bad squares for the queen when h6 a6 considered to be good
>squares for the white queen because it may help in mate attacks in these
>squares).
>
>queen_square_table 990  1000  1000  1000  1000  1000  1000  990 990  1000  1000
>1000  1000  1000  1000  990 1010  1000  1000  1000  1000  1000  1000  1010 1000
>1000  1000  1000  1000  1000  1000  1000 1000  1000  1000  1000  1000  1000
>1000  1000 1000  1000  1000  1000  1000  1000  1000  1000 990  1000  1000  1000
>1000  1000  1000  990 990  1000  1000  1000  1000  1000  1000  990
>
>Uri

Hi Uri,

Yes, my queen piece square table has more amplitude than yours but generally I
try to avoid tucking the queen into corners where its influence is _usually_
going to be reduced.

One other thing though, by raising the value for a6 and h6 you are speculating.
For instance, what if the king is castled q side? and you still give a slight
bonus when the queen is on h6.

Maybe better is some sort of bonus for proximity to the enemy king (which
increases when the enemy king is undefended and/or there are other pieces also
bearing down on the king).

That's what Chrilly is talking about. Its important (if not a little difficult)
to ensure you don't give inappropriate bonuses like in the a6/h6 example above.
Don't worry I make these mistakes and much worse in TRACE... but its an
interesting concept to write an eval that is indeed accurate in the highest
possible % of positions.

Ross








This page took 0.03 seconds to execute

Last modified: Thu, 07 Jul 11 08:48:38 -0700

Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.